HAL Id: hal-00077612
https://hal-insu.archives-ouvertes.fr/hal-00077612
Submitted on 8 Mar 2019
HAL is a multi-disciplinary open access
archive for the deposit and dissemination of
sci-entific research documents, whether they are
pub-lished or not. The documents may come from
teaching and research institutions in France or
abroad, or from public or private research centers.
L’archive ouverte pluridisciplinaire HAL, est
destinée au dépôt et à la diffusion de documents
scientifiques de niveau recherche, publiés ou non,
émanant des établissements d’enseignement et de
recherche français ou étrangers, des laboratoires
publics ou privés.
zone, Chile.
Fidel Costa Rodriguez, Brad Singer
To cite this version:
Fidel Costa Rodriguez, Brad Singer. Evolution of Holocene dacite and compositionnally zoned magma,
volcan San Pedro, Southern volcanic zone, Chile.. Journal of Petrology, Oxford University Press
(OUP), 2002, 43 (8), pp.1571-1593. �10.1093/petrology/43.8.1571�. �hal-00077612�
Evolution of Holocene Dacite and
Compositionally Zoned Magma, Volca´n San
Pedro, Southern Volcanic Zone, Chile
FIDEL COSTA
1AND BRAD SINGER
2∗
1INSTITUT DES SCIENCES DE LA TERRE D’ORLE´ ANS-CNRS, 1A RUE DE LA FE´ROLLERIE, ORLE´ANS 45071, FRANCE 2DEPARTMENT OF GEOLOGY AND GEOPHYSICS, UNIVERSITY OF WISCONSIN–MADISON, 1215 WEST DAYTON
STREET, MADISON WI 53076, USA
RECEIVED AUGUST 6, 2001; REVISED TYPESCRIPT ACCEPTED FEBRUARY 21, 2002
Volca´n San Pedro in the Andean Southern Volcanic Zone (SVZ) Consequently, the role of young, unradiogenic hydrous gabbro in Chile, comprises Holocene basaltic to dacitic lavas with trace generating dacite and contaminating basalt may be underappreciated element and strontium isotope ratios more variable than those of in the SVZ.
most Pleistocene lavas of the underlying Tatara–San Pedro complex. Older Holocene activity built a composite cone of basaltic andesitic and silicic andesitic lavas with trace element ratios distinct from
KEY WORDS:Andes; dacite; gabbro; Holocene; strontium isotopes
those of younger lavas. Collapse of the ancestral volcano triggered the Younger Holocene eruptive phase including a sequence of lava flows zoned from high-K calc-alkaline hornblende–biotite dacite to two-pyroxene andesite. Notably, hornblende–phlogopite gabbroic
INTRODUCTION
xenoliths in the dacitic lava have relatively low 87
Sr/86
Sr ratios
The origin of compositional zoning in magma erupted
identical to their host, whereas abundant quenched basaltic inclusions
within volcanic arcs remains controversial (Eichelberger
are more radiogenic than any silicic lava. The latest volcanism
et al., 2000, 2001; de Silva, 2001). It has been investigated rebuilt the modern 3621 m high summit cone from basaltic andesite
at several localities and scales including relatively large
that is also more radiogenic than the dacitic lavas. We propose
systems such as the >550 km3Calabozos complex
(Grun-the following model for (Grun-the zoned magma: (1) generation of
der & Mahood, 1988) and 50 km3 Crater Lake (Bacon
hornblende–biotite dacite by dehydration partial melting of
phlo-& Druitt, 1988; Druitt phlo-& Bacon, 1989),
intermediate-gopite-bearing rock similar to the gabbroic xenoliths; (2) forceful
sized systems such as the 15 km3
Valley of Ten Thousand
intrusion of basaltic magma into the dacite, producing quenched
Smokes (Hildreth, 1987), 9 km3
Quizapu (Hildreth &
basaltic inclusions and dispersion of olivine and plagioclase
xeno-Drake, 1992) and 4 km3 Giant Crater–Medicine Lake
crysts throughout the dacite; (3) cooling and crystallization–
(Baker et al., 1991), and much smaller 0·1–1·4 km3
erup-differentiation of the basalt to basaltic andesite; (4) mixing of the
tions such as Paricutin, Arenal, and Rishiri (Wilcox,
basaltic andesite with dacite to form a small volume of two-pyroxene
1954; McBirney et al., 1987; Reagan et al., 1987; Kuritani,
hybrid andesite. The modern volcano comprises basaltic andesite
2001). Whereas explosive eruptions like those at Quizapu,
that developed independently from the zoned magma reservoir.
Crater Lake, and Valley of Ten Thousand Smokes can
Evolution of dacitic and andesitic magma during the Holocene and
provide an instantaneous sampling of a zoned magma,
over the past 350 kyr reflects the intrusion of multiple mafic magmas
particularly revealing records occur as well in historical or
that on occasion partially melted or assimilated hydrous gabbro
Holocene lavas and tephras at Medicine Lake, Paricutin,
within the shallow crust. The chemical and isotopic zoning of
Arenal, and Rishiri, where superposition of different
Holocene magma at Volca´n San Pedro is paralleled by that
Moreover, shifts in erupted magma composition over short periods of time may illuminate time scales for both differentiation processes and the invasion of subvolcanic plumbing systems by new batches of magma (Singer et
al., 1995; Hobden et al., 1999).
This paper focuses on a sequence of lava flows of 1 km3
that erupted during the Holocene from a strongly zoned body of magma beneath the composite Volca´n San Pedro in the Southern Volcanic Zone of the Chilean Andes. The lava ranges from biotite–hornblende dacite (66 wt % SiO2) to a two-pyroxene dacite (64 wt % SiO2) that
contains quenched inclusions of basalt (52 wt % SiO2),
andesite (61 wt % SiO2), and basaltic andesite (55–
57 wt % SiO2). Whereas petrographic and mineralogical
evidence for physical mingling of basaltic and dacitic magmas is clear (Singer et al., 1995), the importance of hybridization via magma mixing, crustal assimilation, liquid fractionation, and crystallization remain poorly known. On the basis of a combination of geological mapping, petrography, whole-rock chemical com-positions, and Sr isotope data we provide insight into the physical processes that contributed to the pre-eruptive chemical zoning. Further, the petrographic and geo-chemical character of the zoned late Holocene magma is compared with the preceding eruptions of dacite and andesite during the latest Pleistocene and earlier Holo-cene to highlight the number and diversity of magma
Fig. 1. Geological setting and location of Volca´n San Pedro in the batches and magmatic processes that have occurred over
Southern Volcanic Zone of the Chilean Andes. Paleozoic (PZ) rocks the past>350 kyr. crop out in the Coastal Cordillera and Cordillera Frontal, whereas Volca´n San Pedro grew at the arc front on crust that Mesozoic ( MZ) rocks crop out beneath the axis of the Andean Cor-dillera, which is also locally intruded or overlain by Cenozoic rocks is probably 35–40 km thick along the transect from which
not shown for clarity. The Central Valley comprises Neogene and Hildreth & Moorbath (1988) obtained geochemical and Quaternary sediment overlying Mesozoic basement. Modified from isotopic data that led to the hypothesis of lower-crustal Hildreth & Moorbath (1988), who inferred that the crust thins markedly southward from>55–65 km between 33 and 34°S to >35–40 km Melting, Assimilation, Storage and Homogenization
between 36 and 37°S. ( MASH) for continental arc magmatism. Geochemical
and isotopic surveys by Davidson et al. (1987, 1988)
further indicated that crustal contamination and mixing between two periods of volcanic activity ( Fig. 2a; Table were important processes at Volca´n San Pedro and the 1): the Older Holocene during which a composite cone underlying volcanic complex. By addressing in detail the of >1 km3 grew, and the Younger Holocene during
origin of the dacitic magma, how it came to be juxtaposed which an additional >1 km3
of lavas that post-date with basalt before eruption, and the nature of the mafic collapse of the southeastern flank of the ancestral cone magma that subsequently recharged the plumbing sys- were erupted, including the modern summit cone of tem, we explore subvolcanic processes taking place above Volca´n San Pedro. The latter reflects the most recent the proposed deep crustal MASH zone of Hildreth & volcanic activity that spanned the last 930 kyr at the Moorbath (1988). TSPC (Singer et al., 1997).
Older Holocene lavas are mainly basaltic andesites that erupted from a central vent at>3500 m above sea level and flowed into valleys glacially incised through the
GEOLOGY OF VOLCA
´ N SAN PEDRO
Volca´n Tatara shield ( Fig. 2b). A silicic andesite (62 wt % Volca´n San Pedro is the composite>2 km3edifice cap- SiO2) from this unit was also described by Ferguson et
ping the Quaternary Tatara–San Pedro complex ( TSPC) al. (1992), however. We present new data from three
at 36°S, 71°50′W in the Southern Volcanic Zone (SVZ) basaltic andesitic lava flows and a silicic andesite flow of the Chilean Andes ( Fig. 1). Except for small neoglacial sampled near its head at about 3500 m ( Fig. 2a and b). moraines, San Pedro is unglaciated, and therefore is of Field relations do not reveal whether basaltic andesitic
or the silicic andesitic lavas erupted first. Holocene age. Throughout this paper we will distinguish
Fig. 2. (a) Geological map of Volca´n San Pedro showing localities of samples. (b) View of San Pedro from the south at 1600 m above sea level in Estero Quebrada Hondo. (c) San Pedro from a light aircraft, view to the NW. Field relations of the Younger Holocene lavas are clearly visible.
Younger Holocene volcanism began after south- hornblende-phyric dacitic lava that forms the earliest emplaced head of the composite flow is designated Qcf eastward collapse of the 1500 m tall cone produced a
13 km long,>4 km3debris avalanche deposit that flowed 1 ( Fig. 2a and c). The Qcf 1 flow was buried by a similar,
more voluminous but less silicic lava, Qcf 2, that hosts down Estero Pellado and Rio de la Puente valleys ( Fig.
1a; Ferguson et al., 1992; Singer et al., 1995, 1997). The rare quenched basaltic inclusions and 2–3 vol. % of angular gabbroic xenoliths 5–50 cm in diameter. These ensuing eruption of an airfall tephra and the strongly
zoned composite San Pedro flow may have been triggered xenoliths, described in detail by Costa et al. (2002), comprise both coarse-grained Late Miocene hornblende-by this sector collapse. The small yet distinctive biotite–
Table 1: Eruptive sequence, composition, mineralogy, and main petrological features of the Older and Younger
Holocene lava flows of Volca´n San Pedro; Younger Holocene map units arranged from oldest at bottom to
youngest at top
Map Sample SiO2 Phenocryst minerals Volume Description and comments
unit no. (wt %) (km3)
Younger Holocene
San Pedro summit-forming eruptions
Qsp 2 H-70 56·7 Pl, Ol, Cpx, Opx, Mt 0·1 Latest flows form modern San Pedro summit spatter cone
Qsp 1 PED-12 55·5 Pl, Ol, Cpx, Mt 0·1 Early flow from summit vent
Composite zoned San Pedro flow
Qcf 4 H-20 61·4 Pl, Cpx, Opx, Mt, Ilm 0·1 Latest eruption from pre-summit vents, contains <1%
H-20i 57·3 Pl, Ol, Cpx, Hbl, Mt basaltic andesite inclusions, Ol+ Pl xenocrysts
Qcf 3 H-23 63·7 Pl, Cpx, Opx, Mt, Ilm 0·5 Composite multi-lobate flow, contains 1–20% basaltic
H-8 63·6 Pl, Cpx, Opx, Mt, Ilm inclusions Ol+ Pl xenocrysts; rare Hbl and Bt with
reaction rims
H-23i 52·4 Pl, Ol, Cpx, Opx, Hbl, Mt
H-8i 52·5 Pl, Ol, Cpx, Mt
Qcf 2 H-12 64·5 Pl, Cpx, Opx, Hbl, Bt, Mt, Ilm 0·2 Dacite contains 0·1% quenched basaltic inclusions and
H-13 64·6 Pl, Cpx, Opx, Hbl, Bt, Mt, Ilm 5% angular hornblende–phlogopite-bearing gabbroic
H-11i 51·5 Pl, Ol, Cpx, Hbl, Mt xenoliths
Qcf 1 H-16 65·8 Pl, Cpx, Opx, Hbl, Bt, Mt, Ilm <0·01 Small lobe of dacite, lacks xenoliths, contains Ol+ Pl
xenocrysts
Older Holocene
Qoh H-73 55·4 Pl, Ol, Cpx, Opx, Mt 1·0 Basaltic andesite flows of Estero San Pedro and
QH1-1 54·8 Pl, Ol, Cpx, Mt Quebrada Hondo
QH1-2 53·9 Pl, Ol, Cpx, Mt
H-72 62·2 Pl, Cpx, Opx, Mt <0·05 Andesite flow of upper Estero San Pedro—near modern
summit
Letter ‘i’ in sample number designates a quenched inclusion. Pl, plagioclase; Ol, olivine; Cpx, clinopyroxene; Opx, ortho-pyroxene; Hbl, hornblende; Bt, biotite; Mt, magnetite; Ilm, ilmenite.
and phlogopite-bearing gabbroic rocks and incompletely another. Thus, whereas the eruptions from a common vent system that produced the composite zoned San solidified, possibly comagmatic (Holocene) gabbro. A
larger two-pyroxene dacitic lava, Qcf 3, containing 5–10 Pedro flow may have persisted up to several thousand years, they just as likely lasted no more than a few vol. % of quenched basaltic inclusions up to 20 cm across
( Fig. 3) erupted next, accompanied by mingled airfall centuries or decades.
A basaltic andesitic lava flow, designated Qsp 1, formed tephra of identical composition (Ferguson et al., 1992;
Singer et al., 1995). The Qcf 3 flow was followed by a part of the eastern slope of Volca´n San Pedro and erupted from a now eroded or buried vent high on the eastern small two-pyroxene andesite lava, Qcf 4, that erupted
from a vent that developed into a small scoria-spatter flank of the edifice ( Fig. 2a and c). We interpret the Qsp 1 lava to represent the initiation of volcanism that built cone above the Qcf 2 dacite ( Fig. 2a and c).
These eruptions took place following the last major the modern San Pedro summit cone. The youngest eruptions were of basaltic andesite (>0·1 km3), designated
retreat of ice after 23 ka (Singer et al., 2000), which
preceded the growth and collapse of the ancestral San Qsp 2, that form the modern 3621 m high summit scoria cone and lava flows that are unglaciated ( Fig. 2a and c). Pedro cone (Qoh lavas, Fig. 2). The Qcf lavas are partly
buried by and therefore older than small neoglacial To help guide the petrological and geochemical discussion that follows, the eruptive history and character of the moraines <1000 years old and the modern summit
cone lavas ( Fig. 2). No ash or soil deposits separate Older Holocene and Younger Holocene lavas are sum-marized in Table 1.
Fig. 3. Petrography of dacite and quenched basaltic inclusions in dacites forming the composite San Pedro lava flow. (a) Outcrop near locality H-23 in Fig. 2c. Grey inclusions up to 20 cm across have crenulated margins and form up to 20% of outcrops within the black glassy dacite host lava. (b) Photomicrograph of quenched basaltic inclusion H-23i; plane-polarized light. A few small hornblende crystals and large forsterite and anorthite crystals are surrounded by plagioclase microlites and interstitial brown glass. (c) Photomicrograph of sample H-16 from Qcf 1 dacite; plane-polarized light. Euhedral biotite, hornblende, augite, plagioclase, and magnetite are visible. Larger sieve-cored An84plagioclase
crystals are similar to those in (b) and probably are xenocrysts. (d) Photomicrograph of sample H-23 from Qcf 3 dacite illustrating the rare anhedral and strongly resorbed hornblende crystals with orthopyroxene+ plagioclase + magnetite reaction rims; plane-polarized light. An, anorthite; Bt, biotite; Hbl, hornblende; Fo, forsterite; Cpx, clinopyroxene; Opx, orthopyroxene.
From these observations we infer that: (1) the composite The range of basaltic to dacitic compositions forming the Younger Holocene lavas contrast markedly with San Pedro flow—including in eruptive order flow lobes
Qcf 1, Qcf 2, Qcf 3 and Qcf 4—reflects rapid and nearly >10 km3of basaltic andesitic lavas (52–56 wt % SiO 2) that
formed the younger part of the underlying Volca´n Tatara complete withdrawal of magma from successively deeper
levels of a strongly zoned chamber that contained biotite– shield during latest Pleistocene between about 70 and 20 ka (Davidson et al., 1988; Ferguson et al., 1992; Singer et al., hornblende dacite at its roof, and less silicic two-pyroxene
andesite in its lower reaches; (2) basaltic and dacitic 1997; Dungan et al., 2001). The lowermost portion of Vol-ca´n Tatara was erupted between 100 and 70 ka and is also magma coexisted and mechanically mingled within the
zoned chamber; (3) before the collapse of ancestral San dominantly composed of >10 km3
of basaltic andesitic lavas that are capped by an>1 km3dacite flow that is
Pedro, the wall-rocks of this zoned magma reservoir
comprised, in part, Miocene gabbroic plutonic rocks or distinguished from the Younger Holocene dacite by the absence of petrographic and mineralogical evidence of incompletely solidified gabbro, blocks of which were
incorporated into the magma during the collapse and mingling or mixing with mafic magma ( Fig. 2a; Singer et al., 1995). Although the focus of this paper is on the Younger entrained in rapidly ascending viscous dacitic magma;
(4) the plumbing system was replenished with basaltic Holocene lavas, we place these into an expanded temporal context by providing salient geochemical background on andesitic magma that subsequently erupted to build the
Table 2: Major element, trace element, Sr isotope, and modal compositions of Holocene San Pedro lavas
and quenched magmatic inclusions
Younger Holocene lavas and inclusions
San Pedro composite flow Composite flow-quenched inclusions
Map unit: Qcf 1 Qcf 2 Qcf 2 Qcf 3 Qcf 3 Qcf 4 Qcf 2 Qcf 3 Qcf 3 Qcf 4
Sample: H-16 H-13 H-12 H-23 H-8 H-20 H-11i H-8i H-23i H-20i
wt % SiO2 65·8 64·6 64·5 63·7 63·6 61·4 51·5 52·5 52·4 57·3 TiO2 0·51 0·58 0·60 0·64 0·65 0·74 1·02 0·99 0·99 1·16 Al2O3 16·26 16·44 16·53 16·75 16·78 16·99 18·19 17·35 17·24 17·85 FeO 3·74 4·18 4·29 4·52 4·53 5·24 8·29 8·60 8·48 7·21 MnO 0·08 0·09 0·09 0·09 0·09 0·10 0·15 0·15 0·15 0·15 MgO 1·78 2·16 2·33 2·29 2·35 2·89 5·89 7·37 7·42 3·13 CaO 4·08 4·51 4·59 4·80 4·82 5·54 10·06 8·01 8·07 6·75 Na2O 4·54 4·35 4·31 4·35 4·31 4·27 3·31 3·23 3·27 4·23 K2O 2·70 2·57 2·52 2·46 2·46 2·17 0·88 1·02 1·09 1·62 P2O5 0·16 0·17 0·18 0·19 0·20 0·21 0·17 0·24 0·22 0·32 Total 100·0 100·1 100·4 100·3 100·3 100·2 100·4 100·4 100·3 100·5 Fe2O3 4·16 4·64 4·77 5·02 5·03 5·82 9·21 9·56 9·42 8·01 ppm Rb 95 89 86 88 84 71 21 31 31 43 Cs 4·6 4·2 4·2 3·3 0·8 1·0 1·3 Sr 452 458 461 482 473 504 454 586 564 570 Ba 618 577 562 563 563 525 212 353 330 383 Sc 8·0 9·8 10·7 13·1 25·2 22·1 17·1 V 54 64 73 82 97 214 173 173 124 Cr 15 24 28 23 26 36 148 172 179 b.d. Co 10 12 13 16 41 40 16 Ni 10 18 22 16 18 21 34 90 91 8 Zn 53 56 57 56 59 65 76 78 80 74 Y 12 13 13 14 15 15 17 16 15 17 Zr 163 156 161 167 168 157 78 100 108 118 Nb 5·8 5·7 5·3 5·9 6·0 5·6 2·3 3·8 4·2 4·2 Ta 0·42 0·41 0·37 0·35 0·29 0·24 0·29 Hf 4·00 3·88 4·02 3·77 2·53 2·55 3·05 Th 8·39 7·93 7·39 6·48 1·90 2·52 3·24 Pb 16 16 16 14 14 13 7 9 8 11 La 20·1 19·9 20·0 19·1 12·4 13·0 14·7 Ce 42·8 42·6 42·7 40·7 30·7 26·3 31·8 Nd 17·9 18·5 19·2 19·2 13·3 15·3 Sm 3·32 3·35 3·63 3·73 3·85 3·36 4·08 Eu 0·87 0·88 0·96 0·99 1·24 1·05 1·32 Tb 0·42 0·40 0·45 0·45 0·54 0·41 0·54 Yb 1·28 1·30 1·44 1·37 1·45 1·44 1·57 Lu 0·20 0·20 0·22 0·21 0·21 0·21 0·23 87Sr/86Sr 0·703990 0·704006 0·704031 0·704026 0·704060 2 0·000009 0·000010 0·000010 0·000010 0·000010
Modes (vol. %; phenocrysts >0·3 mm; based on 1500 points, vesicle-free)
Plagioclase 15·3 17·0 19·7 18·0 20·2 25·0 16·6 46·1 29·8 Olivine 0·1 0·6 0·6 0·9 1·0 5·0 7·5 10·2 0·1 Clinopyroxene 0·7 0·3 1·4 0·7 0·9 1·0 1·5 6·2 8·3 Orthopyroxene 0·5 1·0 0·9 1·0 1·2 0·1 2·1 0·1 Hornblende 3·8 1·4 tr tr 5·0 3·7 0·1 Biotite 0·3 0·9 tr tr Magnetite 1·2 0·8 0·5 1·0 0·7 0·5 0·4 2·8 2·4 Sum crystals 21·9 21·9 23·1 21·6 24·0 36·6 26·0 71·2 40·9
Younger Holocene Lavas Older Holocene Lavas
Summit lavas
Map unit: Qsp 1 Qsp 2 Qoh Qoh Qoh Qoh
Sample: PED-12 H-70 H-73 QH1-2 QH2-1 H-72 wt % SiO2 55·5 56·7 55·4 54·8 53·9 62·2 TiO2 1·02 1·01 1·11 1·04 1·02 0·62 Al2O3 18·03 17·70 18·35 18·73 18·03 17·84 FeO∗ 7·47 7·05 7·40 8·15 7·41 4·65 MnO 0·13 0·13 0·14 0·14 0·14 0·10 MgO 4·60 3·77 3·36 3·89 4·29 2·24 CaO 7·63 6·56 7·41 7·73 8·02 5·46 Na2O 3·64 4·17 4·24 4·03 4·17 4·78 K2O 1·36 1·65 1·41 1·36 1·31 1·63 P2O5 0·25 0·26 0·27 0·25 0·39 0·22 Total 100·5 99·7 99·9 100·1 99·5 100·3 Fe2O3 8·30 7·83 8·22 8·15 8·25 5·17 ppm Rb 35 46 34 32 33 38 Cs 1·8 1·4 Sr 556 580 571 576 575 240 Ba 351 430 373 352 319 661 Sc 19·7 17·0 23·0 V 181 142 195 34 178 15 Cr 77 23 9 20 39 b.d. Co 45 24 22 Ni 53 22 14 27 43 4 Zn 75 82 93 77 77 14 Y 18 17 20 19 19 13 Zr 111 147 144 138 136 128 Nb 6·0 5·6 4·1 4·0 4·9 Ta 0·30 0·33 0·26 Hf 3·34 3·67 3·67 Th 3·67 3·07 3·07 Pb 12 10 9 10 10 La 15·4 17·7 16·0 Ce 33·8 42·0 38·4 Nd 17·0 21·0 20·2 Sm 4·01 3·98 4·15 Eu 1·21 1·31 1·40 Tb 0·50 0·56 0·58 Yb 1·55 1·56 2·10 Lu 0·23 0·23 0·29 87Sr/86Sr 0·704072 0·704109 0·704064 0·703952 0·704026 0·703921 2 0·000011 0·000010 0·000011 0·000011 0·000011 0·000011
Modes (vol. %; phenocrysts >0·3 mm; based on 1500 points, vesicle-free)
Plagioclase 26·8 17·7 20·5 26·0 15·4 Olivine 2·0 1·6 0·2 0·7 Clinopyroxene 2·3 0·1 0·1 1·0 0·5 Orthopyroxene 1·4 0·2 tr 0·8 Hornblende Biotite Magnetite 0·1 0·2 0·1 0·1 Sum crystals 31·2 21·0 21·0 27·8 16·8 ∗FeO estimated as 0·9 × Fe2O3.
and c; Singer et al., 1995; Feeley & Dungan, 1996) and lavas at intermediate plagioclase compositions. The high anor-thite population (An82–84) consists of sieve-cored crystals,
forming the Older Holocene portions of the composite
Volca´n San Pedro. with textures and composition identical to those of the quenched basaltic inclusions, and thus these calcic crystals are probably xenocrysts derived from mechanical frag-mentation of the quenched inclusions ( Fig. 3; Singer et
PETROGRAPHY AND MINERALOGY
al., 1995). In addition, numerous glomerophyric clots upOF ERUPTIVE PRODUCTS
to 1 cm diameter with phenocrysts up to 2 mm of calcic plagioclase (An80 cores mantled by An60 rims) + twoThe Older Holocene lavas (samples H-73, QH1-2,
QH2-pyroxenes+ Fe–Ti oxides + glass are present in the 1, H-72; Table 2) exhibit neither macroscopic nor
micro-Qcf 3 dacite. scopic evidence for mingling or mixing, nor is
dis-The dacitic and andesitic lavas contain augitic clino-equilibrium apparent among the phenocrysts. In contrast,
pyroxene (Wo40–46, En41–44, Fs13–16) and orthopyroxene
the Younger Holocene lavas of the composite San Pedro
(Wo2–3, En68–72, Fs26–30); both pyroxenes show slight reverse
flow (samples H-16, H-13, H-12, H-23, H-8, H-20; Table
(>3%) core to rim zoning in enstatite component in the 2), for which we have obtained mineral compositions
Qcf 4 andesite. Fe–Ti oxides include titanomagnetite ( Table 3), contain coarse-grained gabbroic xenoliths
(Costa et al., 2002), quenched basaltic inclusions (samples (XUlv0·23–0·45) and rare ilmenite (XIlm0·80–0·85). Using pairs
H-11i, H-8i, H-23i, H-20i; Table 2), and disequilibrium of adjacent (but not touching) magnetite and ilmenite phenocryst assemblages (Singer et al., 1995; Costa et al., that fulfill the empirical Mg/Mn partitioning test for 2002). Electron microprobe analyses were acquired using equilibrium (Bacon & Hirschmann, 1988) and the so-a JEOL 733 instrument so-at Southern Methodist University lution model of Ghiorso & Sack (1991), we estimated operated at 15 keV and 20 nA with an electron beam of pre-eruptive crystallization temperature and oxygen 5m width. Natural and synthetic mineral standards fugacity for the Qcf 4 andesite and the Qcf 1 and Qcf were used to monitor accuracy and precision, which are 2 dacites. These are >880–990°C near Ni–NiO for >2–3% relative for the major oxides (Singer et al., 1995). the andesite and a slightly lower range of temperature, >840–930°C, but more oxidized, Ni–NiO + 1 log unit, for the dacites [Ni–NiO buffer of Heubner & Sato (1970)]. The large range in calculated crystallization temperature
Older Holocene lavas
for each sample probably reflects the presence of Fe–Ti The three basaltic andesite samples (H-73, QH1-2 and
oxide xenocrysts from the basaltic inclusions. Horn-QH2-1) contain >30% phenocrysts, and have
por-blende- and biotite-bearing lavas are fairly scarce in the phyritic to seriate textures. Plagioclase is the predominant
Southern Volcanic Zone (Hildreth & Moorbath, 1988) phenocryst, followed by clinopyroxene, orthopyroxene,
including the TSPC (Dungan et al., 2001), thus the two and lesser olivine and Fe–Ti oxides ( Table 2). The
most evolved dacite lava flows, Qcf 1 and Qcf 2, are silicic andesite, H-72, has a seriate texture with 20%
unusual because they contain 1–4% of euhedral pheno-phenocrysts of plagioclase, clinopyroxene,
ortho-crysts of both magnesiohastingsitic hornblende (Leake et pyroxene, and minor Fe–Ti oxides.
al., 1997) and biotite. The Al2O3contents of hornblende
vary from 6 to 10 wt % ( Table 3), and the mg-number [= 100MgO/(MgO + FeO) in molecular proportions,
Younger Holocene eruptive products
where total iron is given as Fe2+] from 62 to 71. TheComposite silicic lava flow
mg-number of biotite is 60–66 ( Table 3). Trace amountsof hornblende and biotite rimmed by intergrowths of Dacitic to andesitic lavas forming the composite San
Fe–Ti oxides, orthopyroxene, and plagioclase are also Pedro flow are vitrophyric, with 22–24 vol. % phenocrysts
present in the dacite Qcf 3, which contains up to 20 vol. % ( Table 2) set in pale grey to black glass. Plagioclase is
quenched inclusions of basalt ( Fig. 3). We infer that by far the most abundant mineral either as a phenocryst
breakdown of these hydrous minerals was caused by (15–20 vol. %) or as microlites. Plagioclase compositions
heating of the dacite that occurred before the eruption in the dacitic lavas define distinct modes at An82–84,
of this lava (e.g. Rutherford & Hill, 1993; see below). An55–60, and An45–50 ( Fig. 4), whereas the andesite has
The andesite Qcf 4 does not contain hydrous minerals. also a population of anorthitic plagioclase and a less
well-Experiments on andesitic and dacitic compositions (e.g. defined mode at>An60. Individual plagioclase
pheno-Rutherford & Hill, 1993; Grove et al., 1997; Scaillet & crysts in the silicic lavas typically exhibit repetitive normal
Evans, 1999) indicate that at least 4 wt % H2O is
zoning patterns (An60–45) punctuated by dissolution
sur-necessary to stabilize hornblende, which constrains the faces that are correlated with abrupt shifts in major
pre-eruptive pressure of the dacitic magma to >1·5 kbar. and trace elements (Singer et al., 1995). These multiple
present in all silicic lava flows, but its proportion increases with decreasing silica content of the host lava ( Table 2). These olivines are probably xenocrysts derived from mechanical fragmentation of the quenched basaltic in-clusions (Singer et al., 1995), which are described next.
Quenched basaltic inclusions
Rounded to irregularly shaped inclusions of quenched basalt (e.g. H8i, H23i; Table 2) up to 20 cm across ( Fig. 3), some with vesiculated glassy rims, form 10–20% of various lobes of the Younger Holocene Qcf 3 dacite flow and contain calcic sieve-cored plagioclase, olivine ( Fo82–74),
and augite phenocrysts (Wo39–45, En41–46, Fs12–14) in a matrix
of acicular plagioclase microlites and brown glass (Singer et
al., 1995). A far lesser volume of similar basaltic inclusions
occurs in the Qcf 2 dacite (e.g. H-11i; Table 2). Micro-phenocrysts of orthopyroxene (Wo2–3, En65–73, Fs24–32) and
acicular hornblende are also present in the more crystal-rich inclusions. Plagioclase crystals are bytownite, An80–85,
with thin normal zoning at their rims. Some inclusions contain a small percentage of low anorthite plagioclase (An55–65) displaying textural evidence of reaction by partial
dissolution (e.g. Tsuchiyama, 1985), suggesting that these were incorporated into the mafic inclusion from the dacite host magma. Centimeter-sized, highly vesicular quenched inclusions of basaltic andesite (H-20i; Table 2) within andesite flow Qcf 4 consist of >30 vol. % skeletal plagioclase, minor olivine, augite, magnetite, rare horn-blende, and a matrix of pale brown glass.
Summit-forming basaltic andesitic lavas
The summit-forming basaltic andesite flows, Qsp 1 and Qsp 2 (samples PED-12 and H-70, respectively; Table 2), contain 32% and 21% phenocrysts of plagioclase, minor olivine, Fe–Ti oxides, and two pyroxenes. Clino-pyroxene phenocrysts in flow Qsp 1 are rounded and resorbed (Wo35–44, En43–47, Fs13–20). The latest
summit-forming lava, Qsp 2, contains euhedral augite (Wo42–45,
En41–45, Fs12–15) and orthopyroxene (Wo3, En71–72, Fs24–26).
Most plagioclase crystals consist of a sieved bytownite core (>An80) surrounded by an oscillatory zoned mantle
of labradorite,>An60, producing a bimodal population
distribution ( Fig. 4). The maximum An contents are Fig. 4. Histograms of anorthite content of plagioclase phenocrysts
slightly lower than the calcic xenocrysts of the silicic lavas in Younger Holocene lavas (samples H-16, H-23, H-20), a quenched
basaltic inclusion (H-23i) of the composite flow, and the two summit- or their quenched basaltic inclusions ( Fig. 4). Olivine, forming lavas (PED-12 and H-70). The tri-modal populations (at mainly ranging from Fo
73to Fo68, is less magnesian than
An40–50, An60 and An82) in the andesitic and dacitic lavas of the
that of the quenched basaltic inclusions or xenocrysts in zoned flow should be noted. The anorthite frequency maxima at
the silicic flows. An>83in these lavas match the maximum in the quenched basaltic
inclusion H-23i and we interpret these anorthite-rich plagioclase crystals as xenocrystic and derived from fragmentation of the
quenched inclusions. Similarly, the smaller plagioclase population in
MAJOR ELEMENT, TRACE ELEMENT,
the inclusion H-23i at An55–65 is interpreted as xenocrysts in theAND Sr ISOTOPE COMPOSITIONS
inclusions incorporated from the silicic host lavas. Summit-forming
basaltic andesitic lavas PED-12 and H-70 contain mostly plagioclase
Analytical methods
with frequency maxima at about An80 and An60, i.e. slightly lessMajor and trace element compositions of freshly slabbed calcic than plagioclase in the quenched inclusion H-23i, but almost
Table 3: Representative compositions (wt %) of selected minerals of Younger Holocene San Pedro lavas
and magmatic inclusions
Mineral: clinopyroxene orthopyroxene
Sample: H-16 H-20 H-23i PED-12 H-16 H-20 H-23i
Label: cpx4-1r cpx6-2r cpx1-2r cpx3-11r opx2-1r opx7-1r opx2m
wt % SiO2 50·96 50·62 51·62 52·11 53·56 53·55 53·41 TiO2 0·43 0·68 0·69 0·69 0·36 0·20 0·32 Al2O3 1·80 2·88 2·49 2·10 0·99 1·42 1·88 FeO∗ 8·50 8·82 8·97 8·58 17·09 16·66 16·32 MnO 0·34 0·28 0·24 0·36 0·54 0·41 0·71 MgO 15·00 15·59 15·99 15·69 25·46 25·57 25·82 CaO 21·04 19·09 18·84 19·57 1·36 1·62 1·39 Na2O 0·34 0·36 0·25 0·32 0·09 0·01 0·07 Sum 98·41 98·32 99·09 99·42 99·75 99·44 99·92 Wo 43·1 39·9 39·0 40·4 2·7 3·2 2·7 En 42·8 45·3 46·1 45·1 70·1 70·4 71·0 Fs 14·1 14·8 14·9 14·4 27·2 26·4 26·3
Mineral: hornblende biotite
Sample: H-16 H-16 H-13 H-13 H-16 H-13 H-13
Label: hbl4-2 hbl3-3 hbl1-1 hbl4 bio6-1 bio1 bio1c
wt % SiO2 48·80 44·40 41·78 44·77 37·16 36·48 37·35 TiO2 0·96 1·99 2·90 2·04 3·70 3·72 4·00 Al2O3 5·88 9·71 11·19 9·4 13·57 13·49 13·71 FeO∗ 17·87 12·56 11·78 11·60 15·98 16·32 16·77 MnO 0·65 0·27 0·23 0·22 0·20 0·19 0·19 MgO 13·67 14·19 14·36 14·98 14·30 13·68 14·77 CaO 11·05 11·22 11·17 11·35 0·02 0·02 0·00 Na2O 1·24 2·06 2·44 1·96 0·93 0·92 1·05 K2O 0·34 0·50 0·40 0·52 8·42 8·93 8·95 BaO 0·05 0·04 0·04 0·62 0·54 0·39 F 0·34 0·11 0·05 0·10 0·02 Cl 0·06 0·06 0·07 0·06 F= O 0·14 0·05 0·02 0·04 0·01 Cl= O 0·01 0·01 0·02 0·01 Sum 97·70 97·06 96·32 96·88 95·01 94·32 97·18 mg-no. 62·1 66·8 68·5 69·7 61·5 59·9 61·1
∗Total iron given as Fe2+.
mg-number= 100MgO/(MgO + FeO), in molecular proportions, where total iron is given as Fe2+.
fluorescence (XRF ) at the University of Massachusetts the methods described by Rhodes (1988) and Frey et al. (1990). Splits for XRF were crushed in tungsten carbide, and by instrumental neutron activation analysis (INAA)
precision, and procedural blanks in these laboratories eruption to 0·704060, a value that is higher than any for were established on the basis of replicate measurements the Qcf lavas ( Fig. 6). Younger basaltic andesites that of the BCR-1 standard (Frey et al., 1990). Uncertainties constitute the flank (55·5 wt % SiO2) and summit
in relative weight percentages for major oxides or ppm (56·7 wt % SiO2) of Volca´n San Pedro have87Sr/86Sr of
for trace elements at the 1 level were estimated through 0·704072 and 0·704109, respectively, within the range replicate measurements of BCR-1 and for XRF are: of the quenched inclusions, but higher than values of the SiO2, 0·29; TiO2, 0·25; Al2O3, 0·35; Fe2O3, 0·31; MnO, andesitic and dacitic lavas ( Fig. 6). The composite San
2·93; MgO, 0·30; CaO, 0·23; Na2O, 2·04; K2O, 0·51; Pedro flow and the summit-forming basaltic andesites
P2O5, 1·68; Rb, 3; Sr, 1; Ba, 3; Pb, 3; Zr, 1; Nb, 2; Cr, display collinear trends in major and trace elements ( Figs
3; Ni, 3; Y, 1; V, 2. Similarly, INAA uncertainties are: 5 and 6). In contrast, the quenched basaltic inclusions in Cs, 3; Co, 2; Sc, 2; La, 2; Ce, 3; Nd, 4; Sm, 3; Eu, 2; the Qcf 3 dacite are not collinear with the Younger Tb, 8; Yb, 3; Lu, 4. Holocene lavas for major elements including Al2O3, P2O5,
Sr isotope compositions were determined at the Uni- and MgO ( Fig. 5). It should be noted that the quenched versity of California–Los Angeles using procedures of inclusions (H-20i, Table 2) in the Qcf 4 andesitic lava Feeley & Davidson (1994); nine measurements of the are virtually indistinguishable in major and trace element NBS-987 standard during this study gave a value of composition from the younger summit-forming Qsp 2 0·710264 ± 7 (2 SEM). These, plus replicates of the lava (Figs 5 and 6), suggesting that the later summit-unknown samples, resulted in an external precision of forming basaltic andesite magma resided in the same better than 7 ppm at the 2 level for the Sr ratios in magma reservoir as the silicic magmas that erupted to
Table 2. form the zoned composite flow.
Concentrations of Rb, Zr, La, and Th in general increase with SiO2, whereas Y, Yb, Sr, and Ni decrease
( Fig. 6). Ratios of incompatible elements Zr/Y and La/
Older Holocene lavas
Yb increase linearly with Rb, whereas K/Rb and87Sr/
Older Holocene lavas comprise basaltic andesites con- 86
Sr ratios decrease ( Fig. 7). Although the range of taining 55 ± 1 wt % and a silicic andesitic lava with
incompatible trace element and Sr isotope ratios is small, 62 wt % SiO2 ( Table 2; Fig. 5). Major, minor, and
they vary together in a systematic way that precludes trace element concentrations and elemental ratios of
simple fractional crystallization as the mechanism that the basaltic andesites are similar to those of the
generates the andesitic and dacitic magmas from either Younger Holocene mafic compositions, but with slightly
the quenched inclusions or the summit basaltic andesite. higher Na2O and lower MgO at a given SiO2 content
As we will show below, the compositional and isotopic ( Fig. 5). Older Holocene lavas range in 87Sr/86Sr from
zoning of the silicic lavas are best explained by mixing 0·70392 to 0·70406, with the most SiO2-rich andesite
different proportions of the most evolved dacite with the having the lowest value ( Fig. 6). Moreover, the andesite
summit-forming basaltic andesite. has lower K, Rb, and Sr concentrations and a lower
87Sr/86Sr ratio than the Younger Holocene andesitic
to dacitic lavas ( Figs 5 and 6). Older Holocene lavas
are further distinguished in that they show remarkably
Late Pleistocene Guadal and Tatara dacites
limited ranges of K/Rb and Zr/Y compared with theWhereas rhyolite is exceedingly rare in the TSPC, small Younger Holocene lavas and inclusions, despite having
volumes of dacitic magma (63–68 wt % SiO2) possessing
a larger range of 87Sr/86Sr ratios ( Fig. 7). These trace
contrasting geochemical, mineralogical and textural char-element and Sr isotope differences suggest that Older
acteristics were erupted at various times (Singer et al., and Younger Holocene magmas were derived from
1995, 1997; Dungan et al., 2001). To illustrate these different sources.
contrasts and place the origin of the Younger Holocene San Pedro dacite into a temporal frame of changing subvolcanic processes over the last several hundred
thou-Younger Holocene silicic lavas, quenched
sand years, we briefly describe chemical compositions ofinclusions, and summit basaltic andesites
the Guadal (>350 ka; Feeley & Dungan, 1996; Feeleyet al., 1998) and Tatara (68 ka; Singer et al., 1995) dacites,
The composite San Pedro flow is chemically and
iso-both of which crop out on the flanks of Volca´n San topically zoned from high-K calc-alkaline dacite with
Pedro ( Fig. 2a and c). The Guadal dacites contain 65·8 wt % SiO2 and
87
Sr/86
Sr of 0·703990 to andesite
abundant quenched mafic inclusions, indicating that with 61·4 wt % SiO2and
87
Sr/86
Sr of 0·704026 ( Figs 5
mingling and mixing of magmas were important processes and 6; Table 2). Quenched basaltic inclusions, containing
(Feeley & Dungan, 1996). They contain 66–67·5 wt % 51·5–52·5 wt % SiO2 and 5·9–7·4 wt % MgO, in the
Qcf 3 dacite extend the87Sr/86Sr variation of the zoned SiO
Fig. 5. Major element Harker diagrams (concentrations in wt %) for the Holocene lavas and quenched inclusions. For comparison, Tatara (68 ka) and Guadal (350 ka) dacites are also plotted. In the Na2O vs SiO2diagram the numbers indicate the sequence of eruption of the composite
flow and the lines relate the quenched inclusions to their host lavas.
higher K2O and lower MgO and CaO than the San displays no evidence for mingling or mixing with mafic
magma (Singer et al., 1995). Ferguson et al. (1992) pro-Pedro dacite H-16 ( Fig. 5). Relative to the Younger
Holocene San Pedro dacite, Guadal dacites contain more posed an assimilation–fractional crystallization origin from a basaltic andesite parent magma. The Tatara dacite Rb, La, Th, Zr, and Y, and lower Sr, yet the Sr is slightly
more radiogenic ( Fig. 6). Although Guadal dacites have is collinear with San Pedro major element variations ( Fig. 5) and contains amounts of Ni and Th similar to the K/Rb and Zr/Y ratios remarkably similar to those of
San Pedro dacite, their La/Yb, Sr/Y, and Ba/Y ratios most evolved San Pedro dacite ( Fig. 6). However, the Tatara dacite has much higher La, Y, and Zr, and slightly are much lower ( Fig. 7). Feeley et al. (1998) proposed
that Guadal dacites originated by partial melting of lower Rb and Sr contents than the San Pedro dacites ( Fig. 6). The Tatara dacite also has a lower La/Yb, Sr/ gabbroic crustal rocks at 3–7 kbar and subsequent
crys-tallization and mixing with basaltic andesitic magma. Y, Ba/Y, and Zr/Y, and higher K/Rb ratio than the San Pedro dacites ( Fig. 7). The87
Sr/86
Sr ratio, 0·704060, The contrasts in87
Sr/86
Sr, La/Yb, Sr/Y, and Ba/Y ratios
indicate that the source rocks or processes that produced of the Tatara dacite is indistinguishable from that of the San Pedro dacite, but slightly lower than for the Guadal the Younger Holocene San Pedro dacite were different
from those responsible for the Guadal dacites (see Dis- dacite ( Fig. 6). The mineralogical, textural, trace element, and isotopic differences between these evolved magmas cussion).
In contrast to the Guadal and San Pedro dacites, the point to multiple magma sources and shifts in differ-entiation processes over the past>350 kyr.
Fig. 6. Trace element (concentration in ppm) and Sr isotope variation diagrams for the Holocene lavas, quenched inclusions, and Tatara (68 ka) and Guadal (350 ka) dacites. Arrow in the Ni vs SiO2diagram illustrates fractionation of olivine+ plagioclase + clinopyroxene + magnetite
from basalt to generate basaltic andesite. Dashes lines show mixing trends from the basaltic andesites to the dacites. It should be noted that the silicic Holocene lavas do not fall exactly on the mixing lines because of the effect of mingling, i.e. selective incorporation of forsteritic olivine with high Ni contents.
Introduction). The origin of such zoning has been
DISCUSSION
ascribed to two main processes: (1) crystal fractionation
Origin of compositional zoning in the
including side-wall crystallization whereby less denseYounger Holocene San Pedro magma
evolved liquids migrate towards the upper parts of thechamber
chamber, producing a gravitationally stable stratificationCompositional zoning in silicic systems has been ob- (e.g. Turner & Gustafson, 1981; McBirney et al., 1987; Bacon & Druitt, 1988; (2) incomplete mixing after served in a number of eruptions (see references in the
Fig. 7. Trace element and Sr isotope ratio variation diagrams for the Holocene lavas, quenched inclusions, and gabbroic xenoliths (some xenoliths with extreme trace element ratios are not shown; Costa et al., 2002). For comparison, Tatara (68 ka) and Guadal (350 ka) dacites are also plotted. Continuous line shows mixing trend between basaltic andesite PED-12 and San Pedro dacite H-16 and reproduces most geochemical features of the lavas in the composite San Pedro flow. Arrows delineate models of anhydrous or amphibole-bearing mineral assemblages fractionated from a basaltic andesite parent. Mineral proportions constrained by major element mass balance (see Table 4); diverging arrows in both models reflect a range of partition coefficients (Appendix, Table A1).
intrusion of an unrelated silicic or mafic magma into the of silicic and mafic magmas is likely to be responsible for the major and trace element and Sr isotope zoning of chamber (e.g. Snyder & Tait, 1998; Eichelberger et al.,
2000). The silicic lavas of the Younger Holocene eruption these lavas.
The andesite Qcf 4 contains 61·4% SiO2 and lies
show extensive field, mineralogical and geochemical
T
able
4:
Mass-balance
calculations
for
major
element
fractionation
models
of
Younger
Holocene
San
Pedr
o
lavas
and
quenched
inclusions
Model: (1) Basalt to basaltic andesite (2) Basaltic andesite to dacite (anhydrous) (3) Basaltic andesite to dacite (hydrous) P D C R S.M. (wt %) P D C R S.M. (wt %) C R S.M. (wt %) Sample: H-23i PED-12 PED-12 H-16 SiO 2 52 ·85 5 ·75 5 ·60 ·10 Ol (F o80 )8 ·05 5 ·76 6 ·06 5 ·90 ·17 Ol (F o70 )7 ·96 5 ·90 ·11 O p x 7 ·9 Ti O2 1 ·01 ·01 ·00 ·00 Cpx 3 ·41 ·00 ·50 ·90 ·43 Cpx 5 ·60 ·60 ·11 H b l 1 2 ·4 Al 2 O3 17 ·41 8 ·11 8 ·00 ·07 Pl (An 70 )1 1 ·11 8 ·01 6 ·31 6 ·30 ·03 Pl (An 60 )3 5 ·21 6 ·40 ·06 Pl (An 60 )3 4 ·3 FeO ∗ 8 ·57 ·57 ·40 ·08 Mt 1 ·37 ·53 ·83 ·60 ·16 Mt 3 ·73 ·60 ·14 Mt 3 ·4 MnO 0 ·20 ·10 ·20 ·05 0 ·10 ·10 ·10 ·05 Ap 0 ·40 ·20 ·10 Ap 0 ·4 MgO 7 ·54 ·64 ·60 ·04 4 ·61 ·81 ·80 ·01 0 ·04 CaO 8 ·17 ·77 ·60 ·02 F 23 ·87 ·74 ·14 ·10 ·01 F 52 ·83 ·80 ·25 F 58 ·5 Na 2 O3 ·33 ·73 ·80 ·17 3 ·74 ·64 ·30 ·24 0 ·15 K2 O1 ·11 ·41 ·40 ·05 R 2 0 ·06 1 ·42 ·72 ·80 ·11 R 2 0 ·32 3 ·10 ·34 R 2 0 ·26 P2 O5 0 ·20 ·30 ·30 ·04 0 ·30 ·20 ·20 ·00 0 ·20 ·00 ∗ T otal iron given as Fe 2+ . Model 3 uses the same parent and daughter compositions as model 2. P , parent; D, daughter; C, calculated; S.M., subtracted minerals; F , amount of crystallization; R , residual; R 2, sum of the square of the residuals; Pl, plagioclase; Ol, olivine; Cpx, clinopyroxene; Opx, orthopyroxene; Hbl, hornblende; Mt, magnetite; Ap, apatite.composition between the most silicic dacite Qcf 1 with 65·8% SiO2and the subsequently erupted basaltic
andes-ite Qsp 1 having 55·5% SiO2( Figs 5 and 6), and contains
heterogeneous crystal populations that vary widely in composition, zoning patterns, and textures (e.g. Fig. 4). Its temporal and spatial position strongly suggests that it resided below the dacitic magmas in the zoned chamber and was above the succeeding basaltic andesite of the early summit-forming eruptions. Accordingly, we have tested the hypothesis that mixing played a major role in the compositional zoning observed in the silicic lavas through major element mass-balance calculations. Mix-ing 42% of the basaltic andesite PED-12 and 58% of the most evolved H-16 dacite reproduces the H-20 andes-ite with low residuals (R2 = 0·13; Fig. 8). Modeling of
the trace elements and 87Sr/86Sr ratios using mixing
proportions that fulfill the major element constraints also reproduces the andesitic composition remarkably well ( Figs 7 and 8). The fact that all silicic lavas contain olivine and plagioclase xenocrysts, or quenched basaltic inclusions, different in composition from the
summit-Fig. 8. Major and trace element mixing models for the origin of the forming basaltic andesites, points toward a more complex H-20 andesite. Mixing 42 wt % of basaltic andesite PED-12 and history than the simple mixing scenario proposed above, 58 wt % of the most evolved dacite H-16 produces compositions like
those of the andesite with R2of 0·1.
however. As the highest anorthite and forsterite contents of plagioclase and olivine xenocrysts are like those of the quenched basaltic inclusions, the silicic lavas also preserve
plagioclase (An70), and 1·3 wt % of Fe–Ti oxides, produces
a record of intricate mingling with a magma more
a composition like the basaltic andesite with low residuals magnesian and calcic than the PED-12 composition—a
( Table 4). Upon differentiating to a composition similar magma akin to that forming the quenched basaltic
in-to PED-12, the two magmas reached thermal equi-clusions. The major element composition of these
in-librium, yet remained sufficiently molten to begin mixing clusions, however, precludes this basalt from being the
in the proportions modeled above. The small quantity mafic mixing end-member in the origin of the andesite.
of quenched inclusions of basaltic andesite (H-20i; Table To reconcile these observations, it is useful to consider
2) in the Qcf 4 andesite lava is a tangible vestige of thermo-mechanical relations of the mafic and silicic
end-the mafic end-member that participated in this mixing members (e.g. Bacon & Metz, 1984; Bacon, 1986; Sparks
process. & Marshall, 1986).
Thermal conduction of heat from the basaltic andesite Forceful injection of small quantities of >1100°C
into the dacite would certainly have raised the tem-basalt into voluminous cooler > 900°C dacite magma
perature of the hybrid andesite and possibly devolatilized would quench the basalt in small blobs (e.g. Fig. 3) and
the hybrid magma to conditions outside the stability of would limit the interactions between the two magmas
hornblende and biotite, which are absent in the Qcf 4 (e.g. Bacon, 1986; Sparks & Marshall, 1986; Campbell
andesite and probably completely resorbed. A much & Turner, 1989). The possibility of mixing by mechanical
larger volume of dacite overlying the hybrid andesite blending and chemical diffusion is greatly enhanced if
magma was probably also heated to a lesser degree; in the proportions of mafic and silicic magma are subequal
this part of the chamber hornblende and biotite became and their compositions are such that upon thermal
equi-unstable and although most crystals were dissolved or libration both end-members remain at least partially
reacted to form orthopyroxene+ oxide + plagioclase molten (e.g. Sparks & Marshall, 1986; Oldenberg et al.,
intergrowths, a few relict hornblende crystals survived 1989). We propose that following initial injection of basalt
( Fig. 3d). Moreover, heating of the dacitic magma from into the dacite rapid cooling of the basalt against the
below by the cooling basalt would have promoted thermal dacite led to its crystallization and fractionation to a
buoyancy and possibly convective mixing (e.g. Oldenburg basaltic andesite composition. Using the composition of
et al., 1989). We have proposed that thermal changes
the quenched basaltic inclusion H-23i as parent and the
associated with convective cycling were responsible for PED-12 basaltic andesite as daughter, subtraction of
the repetitive cycles of An45–50plagioclase dissolution and
>24 wt % of a mineral assemblage consisting of 8 wt %
anorthite-rich peaks associated with high Mg, Fe, and unlikely origin for the dacite. In contrast, some features of the Tatara dacite (e.g. Sr/Y, Ba/Y, La/Yb) are Sr concentrations in most of the dacites ( Fig. 4, and see
explained by crystallization of an anhydrous mineral Singer et al., 1995).
assemblage from a parent magma similar to PED-12, although its K/Rb value suggests that some open-system crystallization may have occurred (Ferguson et al., 1992).
Origin of the dacite magma: fractional
To further explore whether the San Pedro dacite can
crystallization vs partial melting of
be derived by fractional crystallization from a mafic
phlogopite-bearing gabbros
parent, we have tested the possibility that hornblendeSeveral petrological and compositional features of the was a crystallizing mineral, as hornblende phenocrysts most evolved dacite Qcf 1 are remarkable in the context are present in all dacitic lavas. Mass-balance calculations of the Holocene lavas and also in the entire TSPC: (1) using the PED-12 basaltic andesite as a parent produce it is among the rare lavas that contain both biotite and compositions like the dacite, and require >58 wt % hornblende phenocrysts; (2) it has higher La/Yb (>15), crystallization of the assemblage 7·9 wt % orthopyroxene, Sr/Y (>36), and Ba/Y (>50) than any other dacite of 12·4 wt % hornblende, 34·3 wt % plagioclase (An
60),
the TSPC; (3) its Sr/Y values are similar to the Holocene 3·4 wt % Fe–Ti oxides, and 0·4 wt % apatite ( Table 4). mafic compositions, but its La/Yb, Ba/Y are much With respect to the dacite, Rayleigh crystallization of this higher, whereas the K/Rb are much lower; (4) its87
Sr/ mineral assemblage produces compositions with lower
86
Sr ratio, 0·70399, is among the lowest of the TSPC. In incompatible element abundances (e.g. Rb), but com-the discussion below we test com-the alternative hypocom-theses parable Sr/Y, Ba/Y, and La/Yb ratios. However, the that this dacite is the product of (1) fractional crys- K/Rb of the calculated composition exceeds the low tallization of mafic magma (plus minor assimilation) or (2) value observed in the dacite, which could be matched partial melting of gabbro. We propose that dehydration only if crystallization was accompanied by assimilation partial melting of phlogopite-bearing gabbroic rocks sim- of low K/Rb material. PED-12 has a K/Rb ratio of ilar to those found as xenoliths in the San Pedro dacite 325, placing it among the lowest of several hundred provides the most plausible explanation for the high Ba/ basaltic and basaltic andesitic lavas measured from within Y and Sr/Y ratios, and low K/Rb and87Sr/86Sr values
the TSPC, of which virtually all that are younger than of the dacite. 350 ka have K/Rb between 600 and 300 (Dungan et al., 2001). Thus, were we to test additional parent magma compositions that might be considered more rep-resentative of the TSPC, we would choose parents with
Fractional crystallization models
K/Rb values much higher than PED-12, none of which Derivation of the dacite from a mafic parent by crystal
are capable of producing the low K/Rb values of the fractionation was tested first by mass-balance calculations
Younger Holocene dacite through crystallization alone. ( Table 4). For these, we used the basaltic andesite
com-Assimilation of silica-rich rocks, or partial melts derived position of PED-12 as parent magma and the mineral
from them, for example the Miocene granitoid plutons compositions determined by electron microprobe.
PED-that form part of basement of the TSPC ( Fig. 2), could 12 was chosen as a parent because it is (1) the most SiO2- lower the K/Rb values significantly, as these granitoids
poor, MgO-rich lava that erupted during the Younger
have high Rb contents (200–280 ppm) and low K/Rb Holocene, (2) similar in composition to basaltic andesite
(140–200; Nelson et al., 1999). However, these plutons that erupted during the preceding Older Holocene period are characterized by87Sr/86Sr values higher than 0·7041
( Figs 5 and 6), and (3) typical of basaltic andesitic lavas, ( Fig. 9a) and thus assimilation would tend to produce which make up the vast majority of the TSPC (Singer et magmas more radiogenic than the dacite. Moreover, the
al., 1997; Dungan et al., 2001). Subtraction of>53 wt % granitoids have lower Ba/Y (<42) and much lower Sr/
of an assemblage comprising >8 wt % olivine (Fo70), Y (<22) so that their assimilation would lower these
5·6 wt % clinopyroxene, 35·2 wt % plagioclase (An60), ratios, producing magmas different from the San Pedro
3·7 wt % Fe–Ti oxides, and 0·4 wt % apatite produced dacite. We next turn to partial melting of hornblende-a liquid resembling the dhornblende-acite (H-16) with low residuhornblende-als and phlogopite-bearing gabbroic rocks akin to the (R2 = 0·3). Rayleigh crystallization trace element
mod-observed xenoliths as a possible origin of the San eling using this assemblage and partition coefficients from Pedro dacite.
the literature (Appendix, Table A1) produces com-positions unlike the dacite; most notably they have lower
Partial melting of hornblende- and
incompatible element abundances (e.g. Rb), much lower
phlogopite-bearing gabbro
Sr/Y, Ba/Y, and La/Yb, and much higher K/Rb ( Fig.
7). Consequently, differentiation of typical mafic magma Partial melting of hydrous gabbroic rocks has been cited as a mechanism to generate silicic magma in a number through crystallization of an anhydrous assemblage is an
Fig. 9. 87Sr/86Sr ratios vs SiO
2contents for lavas, exposed Tertiary basement rocks, and gabbroic xenoliths at Volca´n San Pedro. (a) Comparison
of lavas with local Miocene granitoid plutons, metavolcanic rocks, and hornblende–phlogopite gabbroic xenoliths found in the younger Holocene Qcf 2 dacite. The ranges of87Sr/86Sr ratios of the lavas and the gabbroic xenoliths (0·7039–0·7041) are identical, whereas the values for the
basement rocks are generally >0·7041. Data sources: Guadal dacite, Feeley et al. (1998); Tatara dacite, J. Davidson (unpublished data, 1995); Miocene granitoid plutons and metavolcanic rocks, Nelson et al. (1999); gabbroic xenoliths from San Pedro Qcf 2 lava, Costa (2000). (b) Arrows illustrate changes in the87Sr/86Sr of the most silicic lavas erupted during the four volcanic episodes of the last 350 kyr. QMI, quenched mafic
inclusions.
of settings (e.g. Mount St. Helens dacite, Smith & Lee- compositionally similar to the San Pedro gabbroic xeno-liths. In the dehydration experiments, 20–30% melting man, 1987; Chilliwack Batholith, Tepper et al., 1993;
Guadal dacite, Feeley et al., 1998). Unfortunately, mod- at 900–1000°C produced dacitic to rhyolitic glasses, leaving a residuum of mainly plagioclase (>55 wt %) and eling of partial melting is far less well constrained than
fractional crystallization, in part because one must infer clinopyroxene (>25 wt %). Water-saturated conditions required higher degrees of melting, 30–45%, but lower the assemblage and proportions of residual minerals,
which are rarely sampled. A common approach is to use temperatures, 900–950°C, to produce dacitic to rhyolitic glasses, and a residuum containing >25 wt % horn-the modes produced during experiments that partially
melted mafic rocks (e.g. Beard & Lofgren, 1991; Sen & blende. Trace element modeling using a batch melting equation produces melts with lower concentrations of Dunn, 1994; Wolf & Wyllie, 1994; Rapp & Watson,
1995). The gabbroic xenoliths found in the Younger Rb than the San Pedro dacite, but Sr/Y, Ba/Y, Zr/Y, and K/Rb values that partly overlap it, depending on the Holocene San Pedro dacite flow Qcf 2 (Costa et al., 2002)
are characterized by large amounts of modal hornblende protolith and conditions of melting ( Fig. 10). Calculated dehydration melts from both protoliths have Sr/Y and (to 50%) and phlogopite (to 28%) and display a large
range of trace element abundances and ratios (e.g. Rb= Ba/Y that are more similar to those of the San Pedro dacite than melts produced under water-saturated con-3–80 ppm, K/Rb = 170–630, Ba/Y = 10–70, La/
Yb = 7–12; Fig. 7). Moreover, these xenoliths have ditions, consistent with partial melting under low water contents. In contrast, the melting conditions appear to relatively low 87Sr/86Sr values identical to those of the
San Pedro dacite ( Figs 7 and 9a). We modeled partial have less influence on the Zr/Y and K/Rb values than does the initial composition of the protolith. For example, melting using as protoliths two gabbroic xenoliths
de-scribed by Costa et al. (2002) that comprise variable melts calculated using the Hx14z xenolith as a protolith better explain the high Zr/Y, and particularly the low proportions of modal plagioclase (>25–50%),
horn-blende (>25–50%), orthopyroxene (8–10%), and phlo- K/Rb values of the San Pedro dacite (Fig. 10).
Low K/Rb values relative to most gabbroic xenoliths gopite (>3–4%), and with different abundances and
ratios of trace elements ( Fig. 10). and basaltic to basaltic andesitic lavas of the TSPC are characteristic not only of the San Pedro dacite, but also To constrain the mineral modes and degree of melting
we used the 3 kbar dehydration and water-saturated the older Guadal dacite ( Fig. 10). If these dacites were indeed derived through extensive melting of gabbroic experiments of Beard & Lofgren (1991) on a hornblende
Fig. 10. Trace element ratios vs Rb contents for the Holocene lavas, quenched inclusions, Tatara and Guadal dacites, and gabbroic xenoliths of Costa et al. (2002). Model compositions from the partial melting of two of the gabbroic xenoliths (samples Hx14s and Hx14z, denoted as crosses with filled circles) calculated using the residual mineral modes of dehydration and water-saturated experiments at 3 kbar of Beard & Lofgren (1991; their starting composition 466, charges 113, 158, 144, and 164). The calculated compositional ranges reflect the contrasting extent of melting in the charges, plus the range of possible partition coefficients (Appendix, Table A1).
et al., 1998), their low K/Rb values most probably reflect similar to the xenoliths that contain abundant hornblende and phlogopite, may be an underappreciated, yet im-low K/Rb in the protoliths. In gabbroic xenoliths studied
by Costa et al. (2002), the budgets of K and Rb are portant, mechanism in generating dacitic lavas at the TSPC, particularly those like the Guadal and San Pedro controlled by the proportion of modal phlogopite. For
instance, we have calculated the K/Rb values of phlo- dacites distinguished by low K/Rb and87Sr/86Sr ratios
( Fig. 9a). gopite in the gabbroic xenoliths by using their K2O
concentration measured by electron microprobe (>8·5 wt %; Costa et al., 2002) and assuming that all Rb in the whole rock resides in phlogopite. Mass-balance
The evolving magmatic system beneath
calculations using the three xenoliths that have the highest
Volca´n San Pedro
modal proportions of phlogopite (>10–30%; Costa et
al., 2002) yielded 280–333 ppm Rb in phlogopite, and Eruptions over the past 930 kyr at the TSPC reflect the arrival of many batches of mafic magma in the sub-in turn K/Rb values of 212–256, which overlap those
of the bulk rock (K/Rb = 170–260). Thus, partial volcanic plumbing system that (1) crystallized, mixed, and assimilated crust (Singer et al., 1997; Dungan et al., melting of relatively young (<Miocene) gabbroic rocks,
2001) and (2) occasionally promoted intracrustal melting
SUMMARY AND CONCLUSIONS
(Feeley et al., 1998; this study). Neither the exact sequenceFrom field, petrographic, geochemical, and isotopic of magmas nor details of their genesis have emerged, yet
observations we propose the following model for the this information is essential to constrain rates of melting,
Younger Holocene San Pedro magmas ( Fig. 11): crystallization, mixing, and magma ascent, which in turn
(1) dacite was generated by dehydration partial melting govern eruptive hazards at arc volcanoes (Hobden et al.,
of relatively young gabbroic rocks with unradiogenic Sr 1999; Turner et al., 2000). Mineralogy, trace element
similar to the hornblende–phlogopite gabbroic xenoliths contents, and Sr isotope ratios point to the presence of
found in the dacite. at least six distinct magmas beneath Volca´n San Pedro
(2) Forceful injection of basaltic magma dispersed blobs at various times during the last 350 kyr. The Guadal
of quenched basaltic magma into the lower portion of dacite is a hydrous crustal melt that contained amphibole
the dacitic magma. Mingling and disaggregation of some and quenched inclusions of basalt (Feeley & Dungan,
blobs scattered olivine and plagioclase xenocrysts from 1996) thereby reflecting the coexistence of both basaltic
the basaltic magma throughout the dacite. and silicic magma at 350 ka. In contrast, the 68 ka Tatara
(3) Olivine ( Fo82) and plagioclase (sieve-cored An84)
dacite contains rare resorbed amphibole, no basaltic
phyric basaltic magma ponded as its uppermost surface inclusions, and probably evolved through fractional
crys-solidified beneath a layer of dacitic magma containing tallization of basaltic andesite (Singer et al., 1995). Older
quenched blobs of basalt. The dacite was heated strongly Holocene lavas contain no amphibole, biotite, or basaltic
by the cooling basalt and began to convect. Thermal inclusions, but encompass an unusual range of trace
changes associated with convective cycling caused re-element and87Sr/86Sr ratios that preclude a relationship
petitive cycles of An40 plagioclase dissolution and An60
by fractional crystallization; assimilation of relatively
un-growth, forming oscillatory zoned crystals with high radiogenic crust, as sampled by the gabbroic xenoliths
anorthite peaks corresponding to high Mg, Fe, and Sr ( Fig. 9a), seems likely. We have presented evidence that
concentrations (Singer et al., 1995). Hornblende and Younger Holocene magmatism reflects partial melting
biotite in the lower, hotter portion of the dacitic magma of gabbro to produce amphibole- and biotite-bearing
became thermally unstable and dissolved or reacted with dacitic magma and that mingling and limited mixing of
the melt. the dacite with basalt created a strongly zoned magma
(4) Cooling of the basalt promoted crystallization– body. Subtle differences in trace element and 87Sr/86Sr
differentiation to a basaltic andesitic composition with a ratios ( Fig. 9b) between the Guadal dacite, Older
Holo-temperature close enough to that of the overlying dacitic cene andesite, and San Pedro dacite suggest that
con-magma to permit mixing. Biotite completely dissolved trasting crustal domains or melting behavior participated
and only rare hornblende relicts survived heating in the with each new influx of mafic magma into the subvolcanic
lower portion of the dacite. The upper portion of the plumbing system. Simply stated, there is no evidence that
magma body stayed sufficiently cool that these minerals mafic magma has progressively crystallized or evolved by
remained in equilibrium. Mingling at the interface be-any monotonic process toward compositions richer in
tween the hybrid andesitic magma and the underlying silica, volatiles, or radiogenic Sr over the last 350 kyr
basaltic andesitic magma produced small inclusions of ( Fig. 9b).
quenched, differentiated basaltic andesite within the Generation of zoned magma beneath Volca´n San
andesite. Pedro may have a parallel at Cerro Azul–Quizapu,
(5) Collapse of ancestral Volca´n San Pedro triggered located 45 km to the north of San Pedro ( Fig. 1). The
eruption of the zoned magma body from its top down. 1846 and 1932 eruptions of Volca´n Quizapu were zoned
The modern volcano is built from basaltic andesitic from basalt (52 wt % SiO2) to rhyodacite (70 wt % SiO2);
magma that resided beneath the zoned chamber and the 1846 dacitic lava contained quenched inclusions of
evolved independently from it. basaltic andesite (Hildreth & Drake, 1992). The Quizapu
Magmatism beneath Volca´n San Pedro during the lavas and tephras preserve coherent gradients in trace
Holocene, and over the past 350 kyr, involved several element and87Sr/86Sr ratios similar to the zoned Younger
intrusions of basalt that on occasion induced partial Holocene San Pedro lavas and inclusions, i.e. the most
melting, mixing, or assimilation in the upper crust. The silicic magma was less radiogenic (87
Sr/86
Sr= 0·70389)
compositional zoning of Holocene magma at Volca´n San than the basaltic end-member (0·70404; Hildreth &
Pedro was remarkably similar to that of historic lavas Drake, 1992). Further work will be needed to establish
and tephras at neighboring Volca´n Quizapu. The upper-whether our interpretation of partial melting of young,
crustal contributions to these magmas were not ne-unradiogenic hornblende–phlogopite gabbro in response
cessarily derived from SiO2-rich granitoids. On the
con-to influx of relatively radiogenic basalt may apply con-to
trary, the importance of relatively young, hydrous gabbro Quizapu and possibly other Holocene to Recent eruptions
Fig. 11. Four-stage model for evolution of the zoned magma body that erupted to form the Younger Holocene composite San Pedro lava flow.
Bacon, C. R. & Druitt, T. H. (1988). Compositional evolution of the magma ascending from lower-crustal MASH zones
be-zoned calcalkaline magma chamber of Mt. Mazama, Crater Lake, neath the Andean Southern Volcanic Zone may be
Oregon. Contributions to Mineralogy and Petrology 98, 224–256. widespread.
Bacon, C. R. & Hirschmann, M. M. (1988). Mg/Mn partitioning as a test for equilibrium between coexisting Fe–Ti oxides. American
Mineralogist 73, 57–61.
ACKNOWLEDGEMENTS
Bacon, C. R. & Metz, J. (1984). Magmatic inclusions in rhyolites,We extend our thanks and appreciation to Andrew Wulff, contaminated basalts, and compositional zonation beneath Coso volcanic field, California. Contributions to Mineralogy and Petrology 85, Mike Rhodes, and Fred Frey for providing the major
346–365. and trace element data, and to Jon Davidson and Frank
Baker, M. B., Grove, T. L., Kinzler, R. J., Donnely-Nolan, J. M. & Ramos for determining Sr isotope ratios. Dwight Duering
Wandless, G. A. (1991). Origin of compositional zonation (high-assisted with the electron microprobe analyses. We also
aluminum basalt to basaltic andesite) in the Giant Crater lava field, benefited from discussion and fieldwork with Laurie
Medicine Lake Volcano, Northern California. Journal of Geophysical Brown, Wes Hildreth, Bob Drake, Jon Davidson, Steve Research 96, 21819–21842.
Nelson, Jim Pickens, and Lyn Gualtieri during the course Beard, J. S. & Lofgren, G. F. (1991). Dehydration melting and water-of this project, which was initiated by Michael Dungan. saturated melting of basaltic and andesitic greenstones and am-We are grateful for constructive comments by Alberto phibolites at 1, 3, and 6·9 kb. Journal of Petrology 32, 365–401.
Bindeman, Y. N., Davis, A. M. & Drake, M. J. (1998). Ion microprobe Saal, an anonymous reviewer, and Associate Editor
Den-study of plagioclase–basalt partition experiments at natural con-nis Geist, which helped us to clarify the discussion in
centration levels of trace elements. Geochimica et Cosmochimica Acta 62, several ways. This study was supported by the National
1175–1193. Science Foundations of the USA and Switzerland, and
Campbell, I. H. & Turner, J. S. (1989). Fountains in magma chambers. a Marie Curie Fellowship from the European Community
Journal of Petrology 30, 885–923.
Program: Improving Human Research Potential and the Costa, F. (2000). The petrology and geochemistry of diverse crustal Socio-Economic Knowledge Base, awarded to F.C. xenoliths, Tatara–San Pedro Volcanic Complex, Chilean Andes.
Terre et Environnement 19, 120 pp. (Ph.D. Thesis, University of Geneva).
Costa, F., Dungan, M. A. & Singer, B. S. (2002). Hornblende- and